ML Anthology

A Fixed Size Storage O(n3) Time Complexity Learning Algorithm for Fully Recurrent Continually Running Networks

Jürgen Schmidhuber
NeCo 1992 pp. 243-248
doi:10.1162/NECO.1992.4.2.243 /neco/1992/schmidhuber1992neco-fixed/

Abstract

The real-time recurrent learning (RTRL) algorithm (Robinson and Fallside 1987; Williams and Zipser 1989) requires O(n4) computations per time step, where n is the number of noninput units. I describe a method suited for on-line learning that computes exactly the same gradient and requires fixed-size storage of the same order but has an average time complexity per time step of O(n3).

NeCo Semantic Scholar

Cite

Text

Schmidhuber. "A Fixed Size Storage O(n3) Time Complexity Learning Algorithm for Fully Recurrent Continually Running Networks." Neural Computation, 1992. doi:10.1162/NECO.1992.4.2.243

Markdown

[Schmidhuber. "A Fixed Size Storage O(n3) Time Complexity Learning Algorithm for Fully Recurrent Continually Running Networks." Neural Computation, 1992.](https://mlanthology.org/neco/1992/schmidhuber1992neco-fixed/) doi:10.1162/NECO.1992.4.2.243

BibTeX

@article{schmidhuber1992neco-fixed,
  title     = {{A Fixed Size Storage O(n3) Time Complexity Learning Algorithm for Fully Recurrent Continually Running Networks}},
  author    = {Schmidhuber, Jürgen},
  journal   = {Neural Computation},
  year      = {1992},
  pages     = {243-248},
  doi       = {10.1162/NECO.1992.4.2.243},
  volume    = {4},
  url       = {https://mlanthology.org/neco/1992/schmidhuber1992neco-fixed/}
}